Dear John
Thanks for your kindly help.
I thought the issue of smoothing MRI to match the point-spread function of the SPECT is quite important and I will study and try to improve my analysis based on such.
My current task was to make sure the process I mentioned above could work properly and I could complete the analysis steps recently. However, I found multiple extra-brain voxels of BPM produced images (as following images [5] and [6])
After checking related images, I wonder if such problem related to no skull-stripping was done for SPECT images.
Please see the following images:
[1] SPECT images after DARTEL related process
https://picasaweb.google.com/lh/photo/1IhwZ5MRnGP-gVzULJq9dg?feat=directlink
[2] the [1] image scaled to BPND ones
https://picasaweb.google.com/kcyang2/BPM?feat=directlink#5617728208214909058
[3] MRI GM images after DARTEL
https://picasaweb.google.com/lh/photo/i_XiNKna5QDnAPQb3O0jLw?feat=directlink
[4] DARTEL Template_6
https://picasaweb.google.com/lh/photo/XrDPWLMYsd2v0lLosGd3aQ?feat=directlink
[5] Corr_pos
https://picasaweb.google.com/lh/photo/MJ9uPH4v-I7_lHMgsC4SBg?feat=directlink
[6] Corr_neg
https://picasaweb.google.com/lh/photo/Vo87J24gchepr_xmAQVREQ?feat=directlink
As you can see, [1] and [2] still had skull and some meningeal space that may influence further co-registration process.
I found quite few information regarding skull stripping for functional images (such as SPECT images).
I wonder such was due to this issue was not a problem of the modern co-registration process of segmentated MRI GM images with SPECT images or some technical limitation for doing skull-stripping for SPECT images?
Thanks a lot!
David
|